Peter Wemm wrote:
> On Tue, Sep 2, 2008 at 11:26 AM, Karl Fogel <kfogel_at_red-bean.com> wrote:
>> Branko Čibej <brane_at_xbc.nu> writes:
>>> Indeed. If the default for log and blame is without -g then it stands
>>> to reason that stat and diff should have the same default, doesn't
>>> it. Most users should be aggressively indifferent to svn:mergeinfo
>>> changes. The fact that svn:mergeinfo is built on the generic property
>>> infrastructure doesn't mean that it's a generic property (svn pd
>>> svn:mergeinfo should IMHO require a --force, for example), or that we
>>> should show changes there by default.
>> What he said. Users are already forced to be too aware of the
>> implementation method we chose for recording merge information.
>>
>> -Karl
>
> For what its worth, have a look at 'svn log
> svn://svn.freebsd.org/base/stable/7/sys/nlm'. It was utterly drowned
> by unrelated mergeinfo commits. Roughly 10 of the 200+ commits are
> actual changes.
And that's the type of issue I'm trying to solve. I ran across it with running
log on 1.5.x/CHANGES. Because CHANGES has explicit mergeinfo, any merge to the
branch touches the svn:mergeinfo property, and the signal:noise ratio of log
goes down the toilet.
I'm currently investigating a generic '--ignore-prop' switch, initially for 'svn
st'. It is being written in such a way that we could easily alias
'--ignore-mergeinfo' or '--forget-merges' or
'--dont-get-in-my-way-i'm-trying-to-do-real-work' to mean '--ignore-prop
svn:mergeinfo'. I hope to have a proof-of-concept patch done in a few days.
-Hyrum
Received on 2008-09-02 23:01:33 CEST